5 ms·
It's hard for me to grok this project's design goals. I mean, the basic idea is simple enough to understand: I want to run SQL queries on metadata about my host
by peterwwillis 12y ago
It's hard for me to grok this project's design goals. I mean, the basic idea is simple enough to understand: I want to run SQL queries on metadata about my hosts. I've built and run several different iterations of that same idea, but they didn't require thousands of lines of C++.
The usual implementation is simple: take any host monitor (say, collectd) that can export key/value pairs from a host, or take a log stream over the network and pair it with a host monitor/log scraper to create key/value pairs. Then insert into an SQL engine while appending to a log for a historical record (or PTA/PITR/whatever, i'm not a DBA). Separately you can create a database application to query/modify the database as needed.
But we're talking like, a handful of python scripts that don't ever change except to add new search features. This seems like a big departure from the simplicity of that approach. Am I missing something?
- sanderjd 12y agoFrom what I can tell, the advantage of this is that it doesn't require integrating with an actual database, which then has to be maintained and scaled.
- peterwwillis 12y agoWhen you have tens of thousands of hosts, and this database (because they do use an SQLite backend) is grown exponentially, it doesn't matter what the storage or query engine is. Whatever it is will have to be maintained and scaled.
- sanderjd 12y agoI actually didn't realize it had a sqlite backend, so point taken!